Our full paper on using crowdsourcing and diagramming to support image and video geolocation was accepted for the HCOMP 2017 conference in Québec City, Canada. Only 29% of paper submissions were accepted for this competitive crowdsourcing conference. Congrats to MS Computer Science alumna Rachel Kohler and BS Computer Science alumnus John Purviance, the first and second authors of the paper, respectively.
Here’s the abstract for the paper:
Geolocation, the process of identifying the precise location in the world where a photo or video was taken, is central to many types of investigative work, from debunking fake news posted on social media to locating terrorist training camps. Professional geolocation is often a manual, time-consuming process that involves searching large areas of satellite imagery for potential matches. In this paper, we explore how crowdsourcing can be used to support expert image geolocation. We adapt an expert diagramming technique to overcome spatial reasoning limitations of novice crowds, allowing them to support an expert’s search. In two experiments (n=1080), we found that diagrams work significantly better than ground-level photos and allow crowds to reduce a search area by half before any expert intervention. We also discuss hybrid approaches to complex image analysis combining crowds, experts, and computer vision.
Dr. Luther unveiled our new Civil War Photo Sleuth software to the public for the first time in historic Gettysburg, Pennsylvania. The software uses crowdsourcing and face recognition to identify unknown people in photos from the American Civil War era.
On Friday, Dr. Luther demonstrated the software at an invitation-only event for Civil War photography experts at the Adams County Historical Society. On Saturday and Sunday, he joined Ron Coddington (pictured below), editor and publisher of Military Images magazine, at the 44th Annual GBPA Civil War Artifact and Collectibles Show. We had a table set up showcasing the Civil War Photo Sleuth software and invited collectors to bring their historical photos to us for scanning and real-time analysis and identification. Many took us up on the offer, and by the end of the weekend, Civil War Photo Sleuth had created quite a buzz. More photos of the event are posted on the Military Images Facebook Page.
We look forward to improving the software based on the feedback we received and preparing for a wider release. Meanwhile, anyone interested in beta testing can sign up on a new website for the project, CivilWarPhotoSleuth.com.
Rachel Kohler, a computer science MS student advised by Dr. Luther, successfully defended her master’s thesis today. Rachel conducted interviews with geolocation experts that led to an accepted poster at the upcoming Collective Intelligence 2017 conference. She then led the development of GroundTruth, a software tool that uses crowdsourcing to support expert geolocators. She also conducted several experiments showing that crowds can substantially narrow down an expert’s search space. Congrats Rachel!
Dr. Luther’s panel, titled “The Design, Development and Implementation of Funded Transdisciplinary Digital History Projects: Illustrative Cases of K-16 Collaboration in Action,” was accepted for the 132nd annual meeting of the American Historical Association, to be held January 4-7, 2018, in Washington, D.C. The panel will introduce two funded digital history projects, including Mapping the Fourth of July in the Civil War Era, that enhance the teaching of historical inquiry in K-16 settings. The panelists include Craig Perrier (Fairfax Public Schools), Paul Quigley (Virginia Tech), David Hicks (Virginia Tech), Kelly McPherson (Montgomery County Public Schools), Dr. Luther, and David Cline (Virginia Tech).
We presented demos for five of our projects: Civil War Photo Sleuth, Connect the Dots, GroundTruth, Incite, and Personalized Paths. It was a pleasure to share our work with many VT faculty members, staff, students, and members of the Blacksburg community.
The Crowd Lab fared well at this year’s annual VTURCS (Virginia Tech Undergraduate Research in Computer Science) Spring Symposium. Civil War Photo Sleuth won first place, and Flud won third place, in the Faculty Choice Awards category. Over 30 projects were judged at the symposium, and the awards include a cash prize. Congratulations to Nam, Abby, David, and their graduate student mentors on this accomplishment!
Dr. Luther presented the lab’s research on crowdsourced image geolocation and the GroundTruth project at Carnegie Mellon University. He was an invited speaker for the Crowdsourcing Lunch Seminar hosted by the Human-Computer Interaction Institute (HCII).
Dr. Luther is Co-PI for a new award from the National Endowment for the Humanities (NEH) Humanities Collections and Reference Resources program. The title of the project is “The American Soldier Collaborative Digital Archive.” The one-year, $50,000 award will support planning to expand our Incite software, originally developed for crowdsourced Civil War history, to the new domain of World War 2 history. The PI is Dr. Ed Gitre of Virginia Tech’s History Department.
The American Soldier Collaborative Digital Archive is a project to make available to scholars and to the public a remarkable collection of written reflections on war and the armed forces by American soldiers who fought in the Second World War. In the 1940s, more than 60,000 military personnel responded in their own words to questions posed by the War Department’s new Research Branch about their time in the service. These survey responses were analyzed, summarized, and interpreted by a team of social and behavioral scientists, who after the war reported their findings in a four-volume work, The American Soldier (1949-50). While the Branch’s quantitative data was later digitized and is available through the Roper Center at Cornell University and the National Archives, the responses themselves, the very personal words of thousands of soldiers, have long remained available only to those who could to travel to Washington, DC to read the text on microfilm. Ultimately, we plan to build an online digital archive of approximately 72,000 images of the handwritten survey responses and to provide a tool that will allow these images to be transcribed by students, scholars, and the public so as to render the text searchable. In this way, we propose to quite literally write these tens of thousands of personal expressions of soldiers into the historical record. We are requesting $50,000 from the NEH to fund a yearlong planning process for this project that will enable discussions among experts from multiple organizations in military history, social science research, crowdsourced transcription, digital archiving, and interactive web design. By the end of the grant period year in spring of 2018, we will produce a list of any related primary or secondary material to be linked to or included in the digital archive; a document of technical requirements for the digital archive at its largest scale; a robust data management plan for the image and text data to ensure that it is accessible for the long term; an outreach plan to ensure that the project becomes known to willing transcribers and interested researchers; and a set of agreements concerning the work and other commitments required by involved organizations and individuals. This project seeks to contribute to the NEH’s “Common Ground” initiative and within it the “Standing Together” initiative.